BSTJ 32: 1. January 1953: Frequency Economy in Mobile Radio Bands. (Bullington, Kenneth)

Frequency Economy in Mobile Radio Bands

By KENNETH BULLINGTON

(Manuscript received August 20, 1952)

The various factors affecting ike disability of mobile radio channels are
discussed, and estimates are obtained for the number of usable channels per
megacijde for several present and proposed methods of operation. The lack
of radio-frequency selectivity is the principal harrier to maximum frequency
economy, but this diificuUy can be avoided by sufficient geographical and
operational coordination.

The iiicreaHitig demand for all types of radio services emphasizes the
need for effi(deiit use of the radio frequeiuy spectrum. In mobile radio
operation the numljer of usable channels that can be obtained in the
\'HF and T'HF mobile bands depends iTot only on the width of the in-
dividual channels, hut also on how and where each channel is to be used.
A<^tivity on the same frequency at neighboring locations, and on neigh-
boring freriueneies at the same location both affect the usefulness of a
channel. Halving the channel spacing doubles the number of potential
assignments, but it does not double, and in some cases it does not ap-
preciably ine^rease the number of usable channels.

The usefulness of a single isolated channel is determined by the in-
tensity of its signal above the noise level. Because of the very wide varia-
tion in received signal strength caused by distance, terrain, building
shadounng, etc., the coverage area of a chaiuiel can be distaissed only in
statistical terms. There are likely to be islands of poor signal-to-noise
ratio even (dose to the transmitter, and the coverage gradually fades out
into more spotty conditions at greater distance.

If the same frequency is used at a neighboring location, the familiar
problem of co-channel interference arises. There will now be locations
where the desire."! signal is above noise, but the undesired signal is still
stronger. Thus, the co\'erage area of a channel is reduced by the existence
of the co-channel transmitter; again, it is possible to discuss this reduc-
tion only in statistical terms.

When two channels are being operated on different frequencies in
the same general area, the eo\'erage area of each is limited by sigoal-to-

42

FREQUENCY ECONOMY IN MOBILE RADIO BANDS 43

noise considerations. In addition, each channel may affect the other
because of spurious radiation from transmitters, insufficient receiver se-
lectivity, receiver oscillator radiation, etc. The recent trend toward re-
ceivers with greatly improvetl IF selectivity is worthwhile, but even
infinite IF selectivity cannot solve many of the present interference
problems.

When three or more channels are operating in the same general area,
another type of interference occurs because of intermodulation in trans-
mitters or receivers. If it were technically feasible to build into the equip-
ment sufficient radio frequency selectivity to separate the working chan-
nels, this interference could be removed. In fact, this is not feasible, and
it is necessary to consider possible modulation products from channels
falling within a frequency band several percent wide. The number of
possible interference conditions that result from intermodulation (third
order) rises from 9 for 3 working channels to 50 for 5 (-hanuels, to 450
for 10 chainiels, and to 495,000 for 100 working cliannels. Some of
these interference combinations overlap and fall on the same channel;
but even considering all possible duplication, intei-modnlation inter-
ference rapidly becomes controlling as the number of closely spaced chan-
nels working in the same area is increased.

It is not technically feasible to achieve enough radio frequency selec-
tivity to permit unrestricted and uncoordinated use of many channels
in a given area, unless the channels are, on the average, separated by
about 1 per cent of the operating frequency. For any kind of efficiency
of frequency utilization, it is necessary to have some coordination in the
location of fixed transmitters and in the use of channels. The maximum
efficiency of utilization requires the maximum coordination.

The technical factors that determine channel width, channel spacing,
and the number of usable channels are described and tabulated below.
The first section discusses the principal factors that affect the useful-
ness of channels equipped with transmitters and receivers with perfect
filtering. This is followed by a consitleration of the limitations imposed by
insufficient total filtering and by insufficient radio frequency filtering.
The next sec^tion shows the reduced requirements that are possible by
coordination between systems. Finally, the quantitative data are used
to illustrate the capabilities and efficiencies of various present and pro-
posed methods of mobile system operation.

CHANNELS WITH PERFECT FILTERING

It has been found by experiment that the radio path loss between an-
tennas in a mobile radio system can be ascribed to three principal fac-

->^^

44 THE BELL SYSTEM TECHNICAL JOURNAL, JANUARY 1953

tors: distance, shadowing and standing wave patterns. The variation
with distance from the base station follows the theoretical free space
loss up to 500 feet or more, as long as the points are withm line of sight.
Typical values of the free-space loss are shown in Table I. Beyond
about one-half mile the median path loss over plane earth increases
about 12 db each time the distance is doubled out to distances of 20-30
miles. "^^ ^

In addition to the mcrease in path loss with distance, which is ac-
counted for reasonably well by the theory of radio propagation over
plane earth, bold features of geography such as mountains and large
buildings cause shadow losses that result in irregular coverage patterns.
For example the median loss at street level for random locations in
New York City is about 25 db greater than the plane earth values com-
puted for the distance and antenna heights involved; the corresponding
10 per cent and 90 per cent losses are about 15 and 35 db respectively.^

Superimposed on the above effects which vary relatively slowly with
location are standing wave patterns whose effect on path loss can change
substantially within a foot or so. The standing waves are the result of
random additions of multiple reflections from nearby buildings or ter-
rain, and the variation in path loss follows the Rayleigh distribution for
small changes in distance in urban areas. In other words, there is no
theoretical limit on the deviation from the median but in 1 per cent of
the possible locations the signal is likely to be more than 8 db above the
median value and in 99 per cent of the possible locations the signal level
is not expected to be more than 18 db below the median value.

The motion of the mobile unit through the standing wave patterns
causes signal fluctuations or flutter in the received signal. The flutter

Table I — Free Space Loss Between Dipoles

Separation Between Transmitting and

Free-Space Loss

Receiving Antennas

IjOmc

450 mc

5 ft.
60 ft.

500 ft.
yi mile

16 db
36
56
70

26 db
46
66
80

1 Young W R Jr , Companson of Mobile Radio Transmission at 150, 450, 900
and 3700 Mc. Bell Sys. Tech. Jl., 30, pp. 1068-1085, Nov., 1952.

■i Aikens A J , and L. Y. Lacy, A Teat of 450-Megacycle. Urban Transmission
to a Mobile Receiver. I.R.E., Proc, pp. 1317-1319, Nov., 1950.

3 BuUington, K., Radio Propagation Variations at VHP and UHF. I.R.li;.,
Proc, pp. 27-32, Jan., 1950.

FREQUENCY ECONOMY IN MOBILE RADIO BANDS 45

rate at 150 mc may be as much as 15 cycles per second for a speed of 30
mph and increases as either the radio frequency or the speed of the
mobile unit is increased. The fast acting gain control needed to minimize
the flutter effects is obtained automatically with frequency modulation
but is more difficult to obtain with amplitude modulation. This factor is
cue of the principal advantages of the use of FM instead of AM for
mobile radio systems.

The co-channel mterference to be expected between stations having
equal transmitter powers depends on the path loss statistics for both the
desired and undesired signals. At the edge of the desired coverage area
tliere must be a high probability that the desired signal will be strong
enough to be useful and only a small probability that the undesired
signal will be strong enough to be troublesome. The geographical separa-
tion needed between co-channel stations varies from about four to six
times the desired coverage radius when FM is used and from six to eight
times when AM is used.* If the neetls for mobile channels were uniformly
distributed geographically only a small part of the potential channel
assignments would ever be used in a given area. However, the needs for
mobile channels are usually concentrated in areas of high population
density so that a large percentage of the channel assignments may be
neederl in the same area.

The above estimates on co-channel spacing depend somewhat on the
antenna heights and the type of terrain, and assume that the same fre-
quency is used in both directions of transmission. When the two-fre-
quency method is used with adequate separation between the trans-
mitting and receiving frequencies, the co-channel spacings can be reduced
to about three to five times the coverage radius for FM and to about four
to six times for AM. This reduction of approximately 30 per cent is
possible because the most troublesome interfering path in the single
frequency method (from base transmitter to base receiver) can be elim-
inated in the two-frequency method by suflacient selecti\dty.

The principal reason for using the single frequency method is to pro-
vide communication between two mobile miits when they are relatively
near each other but are beyond the range of the base station. When
transmission of all messages through the base station is desirable, or at
least not objectionable, the two-frequency method is preferable. It is
shown in a later section that close geographical and operational co-
ordination is needed to achieve maximum efficiency in the use of fre-
quency space and that this coordination can be obtained only with the
two-frequency method.

* See reference in Footnote 3.

46 THE BELL SYSTEM TECHNICAL JOURNAL, JANUARY 1953

The bandwidth needed to pass the desired signal depends on the fre-
((uency stability that can be maintained as well as on the type of modula-
tion. The allowance for frequency drift uicludes the variations in both
transmitters and receivers. The importance of these figures is indicated
in Table II which shows the tolerances needed for frequency instability.
For example, with an overall frequency stability of ±0.002 per cent the
channel width at 450 mc needs to be 18 kc wider than the minimum band-
width required to pass the modulated signals.

The use of frequency modulation has several important advantages
that cannot be readily obtained with AM. The instantaneous gain con-
trol and the closer co-channel spacings have already been mentioned.
In addition, for the same radiated power, FM with a frequency swing
greater than about ±3 kc has the well known advantage of providing a
higher output signal-to-noise ratio throughout most of the coverage
area than is possible with double sideband amplitude modulation; this
FM advantage is substantially reduced when the IF bandwidth is large
compared with the bandwidth required to pass the desired sidebands.

The bandwidth required for frequency modulation of a 3 kc voice
band must be at least d=3 kc. For reasonable FM signal-to-noise ad-
vantage, particularly in the presence of impulse noise, the frequency
swing should be at least it5 kc which requires a bandwidth of ±8 kc
for good quality. The corresponding bandwidth for amplitude modula-
tion is ±3 kc; the use of single sideband AM transmission does not seem
feasible, at least not for single channel operation.

LIMITATIONS IMPOSED BY INSUFFICIENT (TOTAl) FILTERING

The frequency separation between carrier frequencies must be greater
than the bandwidth required to pass the desired signal because addi-
tional frequency space or guard bands are needed to build up receiver
selectivity against undesired signals and to avoid the extra band radia-
tion from transmitters. The power of a 100 watt transmitter is about

Table II — Toleeance Needed for Overall Freqxjency

Drift

Frequency Stability

Allowance tor Frequency Drift

150 mc Band

150 mc Band

±0.001%

±0.002

±0.005

±1.6 kc

±3

±7.5

± 4.5 kc
± 9
±22.5

FREQUENCY ECONOMY IN MOBILE RADIO BANDS

47

Tablk in ^ — Required Suppression versus Distance
Between Antennas

Distance Between Transmitting and

Total Selectivity or Filtering Required

Receiving Antennas

150 mc

450 mc

ft.
50 ft.
500 ft.
1^ mile

160 db
124
104
90

160 db
114

94

80

160 db greater than the minimum signal that is useful in the receiver
(140 db below one watt), so ideally no appreciable interference would
result if the overall selectivity of the receiver and the suppression of
extra band radiation in the transmitter could be in excess of 100 db.
This amount of isolation is difficult to obtain by filtering. The inter-
action between transmitter and receiver of the same system is frequently
avoided by the use of "push-to-talk" operation, but the potential inter-
ference between different systems requires the full 160 db (based on 100
watt transmitters). Fortunately, a substantial part of the desired isola-
tion can be obtained by modest geographical separation. The net re-
quirements for either receiver selectivity or transmitter filtering are
less than 100 db by the losses shown in Table I and are summarized in
Table III.

Receiver selectivities of 90-100 db or more are feasible except on
nearby channels and possibly on certain image channels. Typical values
of the guard bands that are required between the edge of the desired
pass band and the frequency at which the desired attenuation to inter-
fering signals can be obtained are estimated in Table IV.

Even if the guard band, shown in Table IV, required to provide ade-
quate selectivity in the receiver could be reduced to zero by providing
infinitely steep sides on the IF selectivity curve, there would still re-
main the guard band needed to avoid the extra band radiation from the
transmitter. The amount of suppression of extra band radiation needed

Table IV — Guard Band versus IF Selectivity

Desired IF Selectivity

Required Guard Band

40 db

60

80

100
120

12 kc
15
20
25

30

■^T*

48 THE BELL SYSTEM TECHNICAL JOURNAL, JANUARY 1953

for unrestricted operation is equal to the required receiver selectivity
given in Table III and can be translated into frequency space in the
following manner.

Both AM and FM transmitters radiate some noise and distortion
products outside of the ideal modulation bandwidth. In addition, some
of the sideband energy in FM falls outside the desired modulation band-
width. The magnitude of the undesired FM sideband radiation is higher
than the noise immediately outside of the desired band, but it decreases
more rapidly with the result that the noise is usually controlling in the
region where the extra band radiation is more than 60 to 70 db down.

The guard bands that are required between the edge of the desired
transmitted band and the frequency at which the necessary suppression
of extra band radiation can be obtained are estunated in Table V.
These values depend on the width of the voice band and are relatively
independent of the radio frequency since r.f. selectivity is not possible.

Measurements on present day transmitters correspond to the above
estimates for values of suppression less than 80 db, but a frequency
separation of nearly one megacycle or more is needed for suppressions of
100 and 120 db. This limitation is not expected to be inherent so more
optimistic estimates are indicated in Table V. If the present charac-
teristics cannot be improved, that is, if suppressions greater than about
80 db cannot be obtained, Table III indicates that some interference
may be expected within about one-half mile of an unwanted transmitter.

A comparison of the information given in Tables III, IV and V in-
dicates that the guard bands required for unrestricted operation are
approximately 100, 50 and 25 kc for minimum separations between
transmitter and receiver of 50 feet, 500 feet and one-half mile, respec-
tively. These values together with the bandwidth needed for modula-
tion and for frequency instability determine the frequency separation
required between channels operating in the same area and are sum-
marized in Table VI.

Table V — Guard Bands Required to Avoid Extra
Band Radiation

Guard Bands Required

Suppression of Extra Band Radiation

AM

FM

40 db

3kc

9 kc

60

10

15

80

25

25

100

50

50

120

100

100

FREQUENCY ECONOMY IN MOBILE RADIO BANDS

49

Table VI — Channel Spacing Required for Unrestricted

Operation of Two FM Channels in Same Area

VERSUS Antenna Separation

Cliannel Spacing, Neglecting Intenaodulation

Minimum Separation Between
Transmitting and Receiving Antenna

ISO mc

450 mc

±0.002%

zhO.0OS%

±0.002%

±0.005%

50 ft.
500 ft.
}4 mile

112-122 kc
62- 72

37- 47

121-131 kc
71- 81

46- 56

124^134 kc
74- 84
49- 59

151-161 kc
101-111
76- 86

The above table shows that if interference of the types so far con-
sidered is to be kept below the minimum usable signal at all distances
greater than about 500 feet from undesired transmitters, the channel
spacing needs to be at least 62 to 75 kc in the 150 mc band and 74 to 105
kc in the 450 mc band. The channel spacings for AM are equal to the
minimum shown above, while the higher figure is for FM with ±5 kc
swing (a modulation bandwidth of ±8 kc).

Since the above channel spacings are considerably greater than the
necessary IF bandwidth, it should be possible to use intermediate chan-
nels in adjacent non-overlapping areas. This geographical Hmitation
does not appreciably decrease the overall efficiency in the use of fre-
quency space as long as the needs for mobile channels are more or less
xuiiformly distributed ^nthin a large region, but it becomes important
where a large percentage of the available channels are needed in the same
metropolitan area.

In a later section it is shown that channel spacings less than the values
given in Table VI are feasible in the same area providing sufficient co-
ordination is achieved in both geographical spacings and operating
methods.

The estimated channel spacings shown in Table VI do not take into
account the effect of intermodulation interference which is discussed in
the follomng section. Intermodulation interference may limit the num-
ber of usable one-way channels to only 1 or 2 per megacycle instead of
the above 6 to 20 per megacycle, unless further restrictions are placed on
the selection of frequencies and on the method of operation.

limitations imposed by insufficient rf filtering

When a strong unwanted signal on a frequency within the RF bandwidth
is present at the input to a receiver, overloading occurs and the receiver

50 THE BELL SYSTEM TECHNICAL JOUBNAL, JANUARY 1953

Table VII — Required RF Receiver Selectivity versus
Antenna Separation

Minimum Separation Between Transmitter

RF Selectivity

and Receiver

150 mc

450 mc

ft. (common antenna)

50 ft.
500 ft.
l-^ mile

95 db
59

39
25

95 db
49
29
15

is said to be desensitized. When two or more strong unwaated signals
are present desensitization also occurs, but in addition, extraneous fre-
quencies are generated by intermodulation in the receiver itself. As the
levels of the unwanted signals become greater than about 75 db below
one watt (1 or 2 millivolts across a typical receiver) the intensity of the
modulation products rises rapidly above the set noise. The resulting
interference can be GO db or more above set noise and the number of
the modulation products increases by at least the cube power of the
number of operating channels.

Ideally, the intermodulation interference ui the receiver caused by
100-watt transmitters (20 db above one watt) can be eliminated by
20 + 75 = 95 db RF selectivity even when the receiver and the unwantf^d
transmitters are connected to the same antenna. In practice, the effect of
geographical separation assuming the free space loss given in Table I
reduces the RF selectivity requirement to the values given in Table VII.

The RF selectivity requirements given in Table VII cannot be ob-
tained on nearby channels. The approximate RF bandwidths associated
with various amounts of RF selectivity in mobile receivers is shown in
Table VIII. For example, in mobile receivers it seems feasible to provide
40 db of RF selectivity at frequencies removed from the desired chan-
nel by about 3 mc in the 150-mc band and by about 10 mc in the 450-mc
band. At fixed stations the RF bandwidth required for a given selec-

Table VIII — Frequency Spacing from Midband versus
RF Selectivity

Desired RF Selectivity

Frequency Spacing from Midband

150 mc

450 mc

20 db

40

60

±1.5 mc
±3

±6

± 5 mc

±10

±20

FREQUENCY ECONOMY IN MOBILE RADIO BANDS

51

Tablic IX — Significant RF Band versus Antenna
Spacing

Minimum Separation Between Receiver and

RF Band

Unwanted Transmitters

150 mc

450 mc

50 ft

500 ft.
\^ mile

±6mc

±3
±2

±14 mc

± 7
± -1

tivity cjan bo reduced to one-third and possibly to one-fourth of the
above values by the use of bulky and expensive filters.

The critical frequency band that needs to be considered in deter-
mining the usefulness of any given channel can be obtained by combining
the information given in the two preceding tables mth the results shown
in Table IX. For example, if it be desired to work mobile receivers un-
restricted to within 500 feet of two or more unwanted transmitters, all
fre<iuency assignments ^Wthin ±3 mc in the 150-mc band (or within
±7 mc hi the 450-mc band) must be carefully chosen if intermodulation
interference is to be avoided.

When the ±3-mc hand is divided into 100 potential channel assign-
ments of 60 kc each and w^hen the chainiels assigned to a given area ai'e
chosen at random, 7 channels working 50 per cent of the time (or 37
channels working 10 per cent of the time) will, on the average, cause
third order intermodulation interference about 10 per <'ent of the time
on each channel within the band. The interference is expected to be
above the minimum usable signal level in all receivers located less than
about a mile from the unwanted transmitters. Even if t he operating fre-
quencies are selected carefully instead of at random, no more than 1 1
channels out of 100 can be found that are free of third order intermodula-
tion when used simultaneously in the same general area. These results
are discussed more completely in a companion paper. '^ When the num-
ber of potential chatuiel assignments is greater or less than 100, the
corresponding number of usable channels limited by thirfl order modula-
tion alone is shown in Table X. The numbers of usable channels shown
above are further reduced when fifth and higher order intermodulation
products are considered.

A reduction in the nominal channel spacing from 60 kc to 20 kc means
a three-fold increase in the potential channel assignments, but Table X
shows that the number of usable channels inerea.ses much more slowly.

^ Bnbcock, W. C, Intermodulation Interference in Radio Systems. Page 63 of
this iaaue.

52

THE BELL SYSTEM TECHNICAL JOURNAL, JANUARY 1953

Table X — Number op Usable Channels versus Number op
Potential Channels

No. of Usable Channels

No. of Potential Channel

Assignments in RF Band

Shown in Table TX

Caretul Selection
No Interference

Random Selection
10% Chance of Interference

% of Time Transmitter Is On

50%

25%

10%

20

50

100

200

500

7

9

11

12

14

5
6
7
9
12

10

12
15
18
24

25
30

37
45
60

Thus far, only the intermodulation interference generated in the re-
ceivers has been considered. Intermodulation also occurs at the same
frequencies in the transmitters, but it usually can be made less impor-
tant than the corresponding interference in the receivers. Ideally, the
intermodulation products generated in the transmitters should not be
stronger than 140 db below one watt (about 1 microvolt at the input to
the receiver) which requires about 75 db RF filtering in each transmitter
output. This ideal requirement is based on 100 watt transmitters with
both the transmitters and recei^'er working on the same antenna. In prac-
tice, the RF filtering requirement is less than 75 db because of physical
separation between transmitters and receivers, and typical values based
on free space transmission are shown in Table XI.

A comparison of the filter requirements on 100 watt transmitters with
the corresponding receiver selectivity requirements given in Table VII
shows that the receiver requirements are greater as long as the effective

Table XI — RF Transmitter Filtering versus Antenna

Separation

RF Filtering Needed in Each Transmitter in db

Distance Between Receiver and
Unwanted Transmitters

ISO mc
Distance Between Transmitters

450 mc
Distance Between Transmitters

ft

10 ft

50 ft

500 ft

ft

10 ft

50 ft

500 ft

Oft.*
50 ft.
500 ft.

3^ mile

75
57
47
40

46
36
29

39
29

22

19

12

75
52
42
35

36
26
19

29
19
12

9
2

• Common antenna.

FREQUENCY ECONOMY IN MOBILE HADIO BANDS 53

separation between transmitters is greater than about 50 feet. For
example, \\ith a 500-foot separation between the transmitting and re-
ceiving anteimas, Table Xll shows that the 150 mc requirement on r.f.
selectivity is 39 db. The bandwidth between the 39 db pomts on the
receiver selectivity characteristic determines the number of potential
channel assignments to be used in Table X.

INCREASED EFFICIENCY OBTAINED BY COORDINATION

The preceding selectivity and filtering requirements are severe and
in some cases vh'tually unattainable except at considerable sacrifice in
frequency space. The principal reason for these exacting requirements is
that the assumed unrestricted and independent operation results in large
<lifferences in field intensities among closely spaced frequencies. In order
to pick out the weak signals from among the strong, sufficient selec-
tivity must be provided to suppress the potential interference to below
the minimum usable signal.

An alternative is to reduce the level differences and hence the filter-
ing requirements by geographical and operational coordination. This
means that the level of the potential interference can be permitted to
be many db above set noise as long as it is always at least 10-20 db
below the desired signal at all possible locations. By proper coordination
the troublesome RF filtering problems can be eliminated within the co-
oi-dinated system and the remaining IF selectivity problems can be
minimized.

The first step is to use the two frequency method of operation with
adequate separation between the frequencies used for the opposite
directions of transmission. In this way substantial EF filtering can be
obtained to eliminate the interference between one or more base trans-
mitters and a base receiver. Tliis type of interference is particularly
troublesome between single frequency systems because of the relatively
high base transmitter jsower and because the high antennas at both
locations reduce the radio path loss to a minimum. The corresponding
possible interference between transmitters and receivers on different
mobile units is also reduced by the two frequency method but inter-
ference between mobile units is much less important because of the
lower power and much lower antenna heights.

The potential interference between base transmitters and mobile re-
ceivers caused by insufficient total filtering can be reduced by locating
all base transmitters at or near a common point so the level differ-
ences between the desired and undesired signals will never be exces-
sive. When all transmitters radiate from a common antenna, a selec-

54 THE BELL SYSTEM TECHNICAL JOURNAL, JANUARY 1953

t'n'ity or filtering re(!uirement of about 40 db (instead of the values
shown in Table III) is sufficient for a reasonable signal to interference
ratio plus an allowance for differential path losses resulting from stand-
ing wave effects.

The IIF selectivity or intermodulation problem in the mobile receiver
can be eliminated by reducing the power level at the first converter to
about 75 db below one watt. This can be done by providing a simple
automatic gain control in the RF stage of the mobile receiver. In regions
where the desii-ed and undesired signals are weak the receiver has full
sensitivity, while at locations near the transmitters both the desired
and undesired signals are reduced in level before reaching the first con-
verter. The result is that the intermodulation products generated in the
receiver are reduced about 3 db for eveiy db that the desired signal is
lowered and the distortion becomes negligible before the output signal-
to-noise ratio is reduced appreciably. In order that the a.g.c. circuit can
l)e fully effective it is necessary that the transmitters be grouped to-
gether and that the desired carrier be transmitted to control the gain of
the receiver.

Grouping the base transmitters at or near a common point together
with the associated measures of transmitting the carriers and using
a.g.c. greatly reduces the requirements on the mobile receiver, but these
measures complicate the design of the base transmitter. The intermodu-
lation products generated in the closely associated transmitters result
in potential hiterfereuce both within and outside of the desired trans-
mitting liand. The intermodulation that falls on the mobile receiver fre-
(luencies needs to be suppressed by at least 25 db below the carrier on
any channel to prevent mutual interference within the coordinated
system. The intermodulation that appears as extra band radiation out-
side the frequency range of the coordinated system must be suppressed
by RF filters. The guard band needed to prevent mutual interference
between the coordinated system and its neighbors is smaU compared
with the frequency space that is saved by the close spacing of the chan-
nels within the coordinated system.

In the direction of transmission from the mobile transmitters to the
base receivers, the above coordinating methods camiot be used but
equally effective ones are available. The RF selectivity requirements
shown in Table VII can be reduced 20 db by using 20 db less power in
the mobile transmitter than in the base transmitter. This measure is
somewhat analogous to the use of a.g.c. in the opposite direction of
transmission ; a further step would be automatic control of the radiated
power but this compHcatiou does not appear to be necessary.

In order to regain the full coverage area, multiple base receivers at

FREQUENCY ECONOMY IN MOBILE RADIO BANDS 55

flilferent locations are needed and this use of space diversity teclmiqiies
provides an opportimity to pick the receiver having the best signal-to-
noise ratio. Moreover, the low power in the mobile transmitter together
with the better RF filters that are possible in fixed locations reduces the
critical bandwidth within which intermodnlation interference can arise
to about ±0.4 mc at 150 mc and to about ±0.0 mc in the 450 mc range.
In these bandwidths approximately 20-25 channels can be obtained
which with random location of the mobile units would be divided more
or less uniformly among five or more base receiving stations. Since no
more than 4 or 5 channels would be operating within the critical RF
bandwidth at any one recei\'ing location, the possibility of intermodula-
tion interference is almost negligible. Finally, an off-channel squelch
circuit is provided which disables the base ^ecei^'^er at a location where
serious adjacent channel interference is most likely to occur and forces
the choice of another base receiver in a different location. Another ef-
fect of the off- channel squelch circuit is that it keeps the base receiver
quiet during idle times, and in this respect it is analagous to the advan-
tage gained in the mobile receiver by continuous transmission of the
desired carrier at the base transmitter.

Most of the above coordiiiating methods tend to emphasize and to
increase the characteristic {lilTerences between the two directions of
transmission. The net effects are that greater frequency economy is ob-
tained and that the electrical requirements are reduced on the mobile
equipment where size, weight and power are critical and where cost
savings are important because of the large number involved. An increase
in complexity occurs at the multi-channel base station but this seems
economically justified because the cost can be divided among many
working channels.

When the above methods of coordination are fully utilized, the RF
re(iuirements are eliminated in the mobile equipment and (^an be met
in the base station equipment. In addition, the IF selectivity require-
ment on nearby chaimels is reduced to about 40 tib in the mobile re-
ceiver and to about 00 db in the base receiver. The extra band radiation
requirement on nearby chamiels is reduced to about 25—40 db in the base
transmitter, depending on whether one or more than one antenna is
used; and to about 00 db in the mobile transmitter.

These requirements coupled with the data given in Tables II, IV
aiid V lead to the frequency separation between coordinated channels
operating in the same area as given in Table XII. Tlie channel spacings
are shown for AM and for FM with a frequency swing of ±5 kc (which
reciuires a bandwidth of ±8 kc for good quality).

The spacings show^n in Table XII assiune that each channel is trans-

56 THE BELL SYSTEM TECHNICAL JOURNAL, JANUARY 1953

Table XII — Channel Spacing versus System Stability-
Coordinated Systems in Same Areas

Ctannel Spacing

stability

150 mc

450 mc

Mobile Receiver

Base Receiver

Mobile Receiver

Bass Receiver

AM

FM

AM

FM

AM

FM

AM

FM

±0.001%
±0.002%
±0.005%

21
24
33

31
34
43

25

28
37

35
38

47

27
36
63

37
46

73

31
40

67

41
50

77

mitted on an individual carrier. Single-channel operation seems to be
the only practical arrangement for transmission from the mobile trans-
mitter to the base receiver. In the other direction of transmission, from
base transmitter to mobile receiver, the question naturally arises whether
additional frequency economy could be achieved by multichannel meth-
ods. In this case individual carrier operation is also indicated for trans-
mission and economic reasons. The multiple echoes that exist at street
level in urban areas limit the number of usable channels that can be
transmitted on a single carrier.^ While the exact number is somewhat
indefinite, it appears to be less than about 20 and perhaps less than 10
channels. In addition the selectivity and liiiearity requirements on
multi-channel receivers (even for two channels) are much more severe
than for single channel equipment. From these considerations it appears
that the use of more expensive receivers and channel separation equip-
ment in each mobile unit is not economically feasible.

FREQUENCY ECONOMY IN PRESENT AND PROPOSED MOBILE SYSTEMS

The technical factors given above provide a basis for estimating the
number of usable mobile channels that can be obtained in a given band-
width. This bandmdth must be sufficiently large to be isolated by RF
filtering if the results are to be well defined.

The following examples assume two different geographical distribu-
tions: (1) the number of usable channels with overlapping coverage
areas that can be obtained within a city or metropolitan area, and (2)
the number of usable channels that can be obtained when the channels
are distributed more or less uniformly over a state or other large area.
The examples are based on the use of frequency modulation with a

■Young W R., Jr., and L. Y. Lacy, Echoes in Transmission at 450 Megacj'cles
from Land-to-Car Radio Units. I.R.E., Proc, pp. 255-258, March, 1950.

FREQUENCY ECONOMY IN MOBILE RADIO BANDS 57

modulation bandwidth of ±8 kc and a frequency stability of ±.002
per cent; with these assumptions, the IF passband should be at least 22
kc in the 150 mc band and 34 kc in the 450 mc band. Narrower band-
widths could be used but this would result in a substantial sacrifice in
coverage under impulse noise conditions.
Five cases are considered:

(1) Single Frequency Semi-Coordinated — In this case, substantially
no hiterference is expected from third order modulation problems, which
are avoided by careful selection of operating frequencies, but higher
order modulation products may be important. Base station locations
are unrelated geographically to other systems in same general area,
except that a minimum spacing of 500 feet between receiver and in-
terfering transmitter is assumed.

(2) Single Frequency with Interjerence—ln this case, the choice of
frequencies is unrestricted, but a 10 per cent chance of third order inter-
modulation interference is accepted within 500 feet of unwanted trans-
mitters, when transmitters are hi operation 25 per cent of time.

(3) Two Frequency Semi-coordinated— This is the same as (1), ex-
cept with two-frequency operation.

(4) Two Frequency with Interference — Same as (2) except with two
frequency operation.

(5) Fully Coordinated Broad-band — This case assumes: (a) two fre-
quency operation with the land transmitters coordinated in location,
power, antenna lieight and emission of protective carriers; (b) low power
mobile transmitters; fc) multiple land receivers; (d) no interference from
third or higher order intermodulation; and (e) guard bands to protect
mobile and neighboring services from mutual interference.

The number of usable channels that can be obtained in the same area
is estimated in Table XIII for frequencies near 150 mc.

The minimum channel spacing shovm in the first column of Table XIII
is calculated as follows: in cases (1), (2), (3) and (4), the extra band
radiation from the base transmitter is controlling. As shown m Tables
III and V, to avoid interference for distances greater than 500 feet from
the interfering transmitter requires a guard band of about 50 kc. This
is added to the 22 kc required IF pass-band of which ±8 kc is allowed
for the FM signal, and ±3 kc for 0.002 per cent system instability.
In (5), the adjacent channel receiver selectivity is controUing: Table IV
shows tlie required GO db can be obtained in 15 kc, which added to the
re(|uircd 22 kc IF band gives approximately 40 kc.

It will be noted from Table VII that the assumption of a separation
of 500 feet between the receiver and the interfering transmitter requires

58 THE BELL SYSTEM TECHNICAL JOURNAL, JANUARY 1953

Table XIII — Usable Channels in City at 150 mc

Method of Operation

(1) Single frequency semi-coordinated.

(2) Single frequency with interference,

(3) Two frequency semi-coordinated...

(4) Two frequency with interference

(5) Fully coordinated broad-band*. . , .

Minimum

(Not Average)

Channel Spacing

in Same Area

Number of Usable
Cbanaels in 6 mc

10

14

5

7
45

* Includes three guard bands of 0.8 mc each to protect mobile and neighbor-
ing services from mutual interference.

about 40 db RF selectivity, and from Table VIII that the 40 db selec-
tivity requires that all frequencies within ±3 mc need to be considered.
With 75 kc channel spacing, there are 80 potential assignments in 6 mc.
Table X indicates that 10 one-way channels can be found that are free
of mutual third order intermodulation interference. If the available
bandwidth were 12 mc the number of interference-free channels would
be doubled.

By the same process from Table X, we derive the number of usable
channels shown for case (2).

For cases (3) and (4), the methods are the same, but the number of
usable channels is reduced to one-half that shown for the single fre-
quency cases.

In the fully coordinated broad-band system (case 5) a usable one-way
channel can be obtained every 40 kc. However, three guard bands total-
ing 2.4 mc are provided to protect both the mobile and neighboring
systems from mutual interference. If the available bandmdth were 12
mc the number of interference-free channels would be increased from 45
to 120 smce no additional guard bands would be required.

The comparison between various methods of operation given in Table
XIII applies to 150-mc channels operating in the same city. When the
channels are distributed more or less uniformly over a large area, the
number of usable channels is increased by several factors. The separation
between carrier frequencies in non-overlapping areas needs to be only
slightly greater than the IF pass-band of the receiver, say, 30 kc at 150
mc The guard bands needed in one location can be used in other areas
at geographical separations less than co-channel spacings. Finally the
required geographical separation between co-channel stations is less
for the two frequency method than for the single frequency method and
is less for FM than it would be for AM.

An esthnate of the maximum number of usable channels within a large

FREQUENCY ECONOMY IN MOBILE HADIO BANDS 59

Table XIV — Usable Channels in State or Large Area

AT 150 MC

Method of Operation

Minimum Channel
Spacing*

Number of Usable
Channels in 6 mc

(1) Single frequency semi-coordinated

25 kc

25

25

25

25

108
171

108

171

240

* Assumes adjacent channels are not assigned in same area.

area can be obtained by considering an area whose radius is about six
times the coverage radius of the individual transmitter. A larger area
is unnecessary because single frequency FM channel assignments can be
repeated at this distance, while a smaller area would tend to approach
the common area concept used above. The large area can be di^dded into
9 subareas, each of which can be treated in the manner used in Table
XIII. The results are sho^vn in Table XIV, which again assumes an
FM modulation bandwidth of ±8 kc and ±0.002 per cent overall sys-
tem frequency stability.

The entries hi Table XIV are calculated as follows: Once again, the
smallest band to be considered is limited by the RF selectivity in mobile
receivers to Ci rac; with 25 kc as the minimum chaimel spacing, there are
6000/25 = 240 potential assignments. From Table X, only 12 can be
found to be free of third order intermodulation. With 12 channels in
each of 9 subareas, there is a grand total of 108 channels usable in the
state or large area. With more frequency space, the usable number is
increased in proportion.

By the same process, from Table X we derive the number shown for
case (2).

In the two frequency cases, the co-channel separation can be made
smaller than in the siugle frequency cases, since the most troublesome
case of interference (that between base transmitters and base receivers)
is eased by RF selectivity. Thus, the co-channel separation needs to be
only about 0.7 that for single frequency operation, which means that
there are now effectively 18 instead of 9 subareas. It follows that the
grand total of usable channels is the same in cases (1) and (3) and cases
(2) and (4).

In considering case (5), we note from Table XIII that 40 kc is the
minimum channel spacing usable in a single subarea. However, the larg-
est grand total of channels is found by using 50 kc spacing in the sub-
areas, and assigning the adjacent 25 kc channels to other subareas.

60

THE BELL SYSTEM TECHNICAL JOTJKNAL, JANUABY 1953

Table XV — Usable Channels in City at 450 mc

Method of Operation

Minimum

(Not Average)
Chancel Spacing

Number of Usable
Channels in 14 mc

85 kc

85

85

85

50

12

17

6

9

68

Includes three guard bands of 2.4 mc each to protect mobile and neighboring
services from mutual interference.

Similarly, the guard bands of one subarea can be used for channels else-
where so all of the available 240 channels can be used.

The examples given in Tables XIII and XIV represent the two
extreme conditions and the practical situation lies in between the two.

By similar reasoning it is possible to estimate the number of usable
channels that can be obtained at frequencies around 450 mc. The num-
ber of usable channels shown in Table XV is for overlapping coverage
areas in a city or metropolitan area and the estimates given in Table
XVI are based on a uniform distribution over a state or other large sec-
tion of the country.

Again, FM modulation with a bandwidth of =t8 kc and a system
frequency stability of 0.002 per cent are assumed.

For a bandwidth of 28 mc instead of 14 mc the number of usable
channels is doubled for the first four cases and is increased from 68 to
208 for the fifth case. The corresponding estimates for bandwidths less
than 14 mc are indefinite because of insufficient r.f. selectivity.

CONCLUSIONS

The principal conclusions that result from Tables XIII, XIV, XV
and XVI, and from the precedmg discussion can be summarized as fol-

Table XVI — Usable Channels in State or Large Area

at 450 MC

Method of Operation

Minimum Channel
Spacing*

Number of Usable
Channels in 14 mc

35 kc

35

35

35

35

117

198

117

198

400

* Assumes adjacent channels are not assigned in same area.

FREQUENCY ECONOMY IN MOBILE RADIO BANDS 61

lows:

1. A fully coordinated system requires a band of several megacycles
that can be treated as a unit, but it offers substantial overall frequency
economy and freedom from interference that can be obtained in no other
way. This is particularly true in large metropolitan areas where the
demand is greatest. With the same equipment and the same standards
of quality and I'eliability, coordinated channels can always be spaced
much closer in frequency than uncoordinated systems.

2. The advantages of coordination increase rapidly as the number of
channels per unit area is increased. However, in areas where only three
or four channels are required, the advantages of complete coordination
are sufficiently small that only the semi-coordination of careful frequency
allocation is required to preserve overall frequency economy.

3. For maximum economy, where full coordination is not used, the
chaniiels should be assigned as in FM and TV^ broadcasting first to areas
and then to users within areas. The allocation of a block of channels to
a particular service with a minimum of operational and geographical
restriction frequently results in an ever-increasing interference problem
as each additional station is placed in operation.

4. Smgle-frequency operation is most suitable where the operational
need for single channel communication between mobile units (as con-
trasted mth hxed-to mobile) is more important than frequency econ-
omy.

5. A frequency separation between potential channel assignments of
25 kc in the 150 mc range, and 35 kc in the 450-mc range seems tech-
nically feasible; but adjacent channels with these minimum spacings
cannot be assigned in the same area. These values may be reduced to
about 20 and 30 kc, respectively, at the sacrifice of an appreciable reduc-
tion in coverage under impulse noise conditions. A further reduction in
channel spacing would not appreciably increase the total number of
usable chainiels, since the controlling factors are RF selectivity and
extra band radiation, rather than IF selectivity or the total number
of potential channel assignments.

6. The average spacing needed between channels operating in the
same area varies from about 40 to 500 kc or more, depending on the
method of operation and the criterion of usability.

7. The need is for a certain small number of channels in all areas,
plus a peaked demand in centers of population. In the semi-coordinated
cases, the maximum number of channels that can be allocated to the
peak area is a small fraction of the total number of channels available.
In the fully coordinated, broad-band case, there is much more flexibil-

62 THE BELL SYSTEM TECHNICAL JOUBNAL, JANUARY 1953

ity and the peak area can be allocated a large fraction of the total avail-
able.

8- FM is preferable to AM for land mobile service because its instan-
taneous gam control feature minimizes the flutter caused by the motion
of the mobile unit through standing wave patterns. This advantage in-
creases in importance as the carrier fre(iuency increases. In addition,
FM with an adequate frequency swing provides an increased signal-to-
noise advantage over most of the coverage area. The somewhat greater
channel width required by FM is more than offset on an area coverage
basis by the closer co-channel spacing.